From Torsional Towards Flexible 6 DOF Models for Dynamic Analysis of Wind Turbine Gearboxes

1
Wednesday 18th March 2009 Marseille European
Wind Energy Conference
From Torsional Towards Flexible 6 DOF Models for
Dynamic Analysis of Wind Turbine Gearboxes
ir. Ben MarrantHansen Transmissions
International NVbmarrant_at_hansentransmissions.com
2
Overview
OVERVIEW

• Introduction
• Purely torsional multibody models
• 6 DOF multibody models with discrete flexibility
• 6 DOF full flexible models
• Status and planning
• Conclusions

3
ORGANISATION
COMPANY STRUCTURE
BE
BE
AU
SA
US
IN
UK
CN
CN
BR
4
PRODUCTS
Hansen helical wind turbine gearboxes

• 1979 30 kW - NF21 Standard industrial gearbox
• 1980/85 30 to 90 kW Standard industrial
  gearboxes
• 1985/92 75 to 250 kW Gearboxes with monobloc
  housing
• 1993 500/600 kW
• Planetary gearboxes
• 1995 1,650/1,750 kW
• 1997 660 and 750 kW
• 2001 850 kW
• Current range 1,5 4,5 MW, developing 6 MW

5
Products wind energy
PRODUCTS
6
Traditional wind turbine design codes
INTRODUCTION INDUSTRY STANDARD

• Flex
• Bladed

7
Traditional wind turbine design codes
INTRODUCTION INDUSTRY STANDARD

• Design codes generate timeseries
• Oversimplified drivetrain
• No external dynamic loading on drive train
  components taken into account
• Overlap of internal eigenfrequencies with
  internal excitations not taken into account
• Need for advanced and validated multibody models
  for accurate dynamic load prediction in multi
  megawatt wind turbine gearboxes

8
Need for multibody simulation in load prediction
INTRODUCTION INDUSTRY STANDARD

• Advanced and validated drivetrain multibody
  models as a basis for more accurate load
  prediction

Torsional model 6 DOF multibody model with discrete flexibility 6 DOF full flexible multibody model
Torque information X X X
Detailed reaction forces X X
Internal component deformations X
9
Modelling approach - DRESP
PURELY TORSIONAL MULTIBODY MODELLING

• Inertias
• Shafts and planet carrier (incl bearings)
• Gears
• Stiffnesses
• Gears
• DIN 3990
• planet ring deformations
• Shafts and planet carrier
• Torsional stiffness
• Bending stiffness (radial tangential gear
  displacement)
• Bearing
• Radial stiffness (radial tangential gear
  displacement)

10
Torsional resonance analysis
PURELY TORSIONAL MULTIBODY MODELLING

• Campbell diagrams show possible resonance areas.
• Excitations from
• unbalance of shafts
• gears
• bearings

11
Torsional resonance analysis
PURELY TORSIONAL MULTIBODY MODELLING

• Sensitivities of eigenfrequencies w.r.t. design
  parameters
• gt more insight in gearbox internal dynamics

12
Measurement campaign
PURELY TORSIONAL MULTIBODY MODELLING
13
Measurement results
PURELY TORSIONAL MULTIBODY MODELLING

• Eigenfrequency identification using Campbell
  diagrams

14
Measurement results
PURELY TORSIONAL MULTIBODY MODELLING

• Overview of identified eigenfrequencies

Frequency (Hz) Description
2.25 Torsional rotational mode anti symmetric around the coupling between GBX1 and GBX2 (this mode is mainly determined by the test rig dimensions)
0-100 HSS axial mode shape
0-100 Torsional rotational mode symmetric around the coupling. Kinetic and potential energy on high-speed side
100-200 Kinetic energy at low speed side and axial displacements of ISS LSS
100-200 Kinetic energy at low speed side
600-700 Kinetic energy at axial displacement of HSS and ISS as well as potential energy at HSS
700-800 Kinetic energy at ISS shaft
15
Model validation and update
PURELY TORSIONAL MULTIBODY MODELLING
Match condition Measured frequency (Hz) Difference () Probability of match Description
DRESP match 2.25 16 0 Very high Torsional rotational mode anti symmetric around the coupling between GBX1 and GBX2
axial mode 0-100 - - HSS axial mode shape
DRESP match 0-100 - 0.3 -1.7 Very high Torsional rotational mode, symmetric around the coupling. Kinetic and potential energy on high-speed side
axial mode 100-200 - - Kinetic energy at low speed side and axial displacements of ISS LSS
DRESP match 100-200 37 0 High Kinetic energy at low speed side
axial mode 600-700 - - Kinetic energy at axial displacement of HSS and ISS as well as potential energy at HSS
no match 700-800 - - Kinetic energy at ISS shaft
16
Conclusions
PURELY TORSIONAL MULTIBODY MODELLING

• Only torsional modeshapes and frequencies can be
  predicted.
• Test showed 3 axial modes could not be predicted.
• According to literature also tilt modes exist
• Flexible gearbox components (housing, planet
  carrier) not represented properly
• System non-linearities not included (load
  dependency, clearance, temperature dependancy,
  ...)
• Not detailed enough for accurate load prediction
• More elaborate (flexible) multibody models
  including 6 DOFs per body implemented in
  Simpack, VL Motion.

17
Modelling approach
6 DOF MULTIBODY MODELS WITH DISCRETE FLEXIBILITY

• Stiffnesses
• Gear contact
• stiffness based on DIN 3990 - ISO 6336
• variable mesh stiffness and clearance
• Shafts and planet carrier
• Torsional stiffness
• Bending stiffness
• Planet ring deformations
• torsional stiffness
• Bearing
• Axial radial stiffness (6x6 matrix)

17
18
Multibody simulation technique
6 DOF MULTIBODY MODELS WITH DISCRETE FLEXIBILITY
19
6 DOF FULL FLEXIBLE MULTIBODY MODELS
Modelling approach

• Finite element models of components in multibody
  simulation using CMS
• Flexible multipoint constraints for introduction
  of discrete forces in flexible structures
• Flexible components
• shafts
• housing
• planet carrier

19
20
STATUS AND PLANNING

• Status
• Template models of all different kind of gear
  stages available
• Implementation of two gearbox models in 13.2MW
  test rig set up including test rig controller
• Planning
• Experimental validation and model updating
• Refining models (adding flexibility, advanced
  bearing model,...)

21
CONCLUSIONS

• Need for advanced and validated multibody models
  for accurate dynamic load prediction
• Approach
• Torsional multibody gearboxes models can be used
  in early design stages
• 6 DOF multibody gearbox models give more insight
  in internal loads and dynamics
• Full flexible multibody gearbox models
  realistically include effect of component
  flexibility
• Need for experimental validation
• 1st step taken gt validated torsional models
• next step gt validation on 13MW test rig

22
(No Transcript)